A Novel Multi-target Detecting and Tracing Method for Robot Vision System

In this paper we study on multi-target detecting and tracing system for intelligent robot and give a good method to detect and track the targets such as pedestrians, which needs to extract the moving targets from the background in image sequence, and track them by probability statistics and controlling theory. In the first stage, we segment the foreground and background regions through GMM (Gaussian Mixture Model) [1] algorithm. Based on the results of the first stage, we build the tracing system with Kalman filter and Mean-Shift in order to capture the moving targets. At last, the experimental results show that our method is correct and robust, it lays a solid foundation for further study on target recognition.